The organization of Golgi in Drosophila bristles requires microtubule motor protein function and a properly organized microtubule array

In the present report, we used highly elongated Drosophila bristle cells to dissect the role of dynein heavy chain (Dhc64C) in Golgi organization. We demonstrated that whereas in the bristle "somal" region Golgi units are composed of cis-, medial, and trans-Golgi compartments (“complete Golgi”), the bristle shaft contains Golgi satellites that lack the trans-Golgi compartment (hereafter referred to as “incomplete Golgi”) and which are static and localized at the base area. However, in Dhc64C mutants, the entire bristle shaft was filled with complete Golgi units containing ectopic trans-Golgi components. To further understand Golgi bristle organization, we tested the roles of microtubule (MT) polarity and the Dhc-opposing motor, kinesin heavy chain (Khc). For our surprise, we found that in Khc and Ik2Dominant-negative (DN) flies in which the polarized organization of MTs is affected, the bristle shaft was filled with complete Golgi, similarly to what is seen in Dhc64C flies. Thus, we demonstrated that MTs and the motor proteins Dhc and Khc are required for bristle Golgi organization. However, the fact that both Dhc64C and Khc flies showed similar Golgi defects calls for an additional work to elucidate the molecular mechanism describing why these factors are required for bristle Golgi organization.


Introduction
The Golgi complex possesses a highly ordered and characteristic morphology. In most mammal cells, the organelle appears as stacks of flattened cisternae that are linked by tubules to give rise to continuous structures known as ribbons [1,2]. In most cell types, the Golgi ribbon is typically located in the perinuclear region near the centrosome. The exact localization is dependent on the microtubule (MT) and actin cytoskeleton, together with active transport or anchoring by motor proteins, such as dynein and kinesin [3]. It was, moreover, shown that MTs are closely associated with the cis-Golgi compartment [4]. In contrast, in non-neuronal Drosophila tissues, the Golgi apparatus appears as stacks dispersed in the cell cytoplasm, juxtaposed to the endoplasmic reticulum exit site (ERES), and does not form the continuous tubular organelle seen in mammalian cells [5]. PLOS

Organelle tracking and statistical analysis
To measure the area and density of MnII-GFP and GalNacT2-YFP-positive particles, confocal Z-stack images of maximal quality of bristle cells were obtained. A Z-stack projection of the bristle shaft was divided into two halves exactly in the middle of the shaft length, resulting in two sections of even length, with one part being close to the base (soma) and the other being distal to the base, referred to as the tip. In the case of a dynein-mutated background, the bristle appeared as two independent parts because of an extremely high signal at the tip area, relative to the base. Images were analyzed using ImageJ software, and the area of each particle was measured automatically using the "analyze particles" tool. The density of MnII and GalNacT2positive particles was measured as the number of particles in each half of the bristle divided by the overall area of that half-bristle (i.e., the number of particles/area of the base or tip half). Statistical analysis of both parameters, i.e., particle area and density, was performed using spiltplot ANOVA with the cell part area being considered as the within-plot treatment and the genotype being considered as the whole-plot treatment. Repeats (random factors) were nested within a genotype. Statistical analyses were performed using STATISTICA, version 10.

Golgi organization in the Drosophila bristle cell
To study the organization of the Golgi in polarized cells, we used the highly elongated Drosophila bristle cell as a model system, together with a set of ERES and Golgi markers. To identify the ERES, we used either Stenosis (Sten) tagged with GFP [24] or antibody staining of Sec16 [26]. As a medial Golgi compartment marker, we followed α-mannosidase II tagged with EGFP (MnII-GFP), while RFP/GFP-tagged N-acetylgalactosaminyltransferase 2 (Gal-NacT2-TagRFP/GFP) and RFP-tagged Gal-T [23] served as a trans-Golgi marker. For visualization of the cis-Golgi compartment, we followed anti-dGM130 antibody staining [23]. Examination of bristle cell Golgi organization revealed that the somal region contains Golgi stacks composed of cis- (Fig 1A' and 1A''), medial- (Fig 1A-1C, 1A'', 1B'' and 1C''), trans-Golgi compartments (Fig 1B and 1B') and ERES ( Fig 1C' and 1C'') scattered throughout the cytoplasm. This type of Golgi is henceforth designated as a complete Golgi unit. Thus, similar to other Drosophila tissues [5,[27][28][29], the soma of the bristle cell contains Golgi mini-stacks and does not create higher-order structures, such as the ring-shaped organization observed in the soma of larval neurons [23]. In Drosophila larval dendritic shafts, the Golgi exists as a discrete "single compartment Golgi" or as a "Golgi mini-stack" composed of more than one compartment [23]. Examination of Golgi organization within the bristle shaft using MnII-GFP as a medial-Golgi marker revealed that the shaft contains MnII-positive particles, these particles are called hereafter Golgi satellites, (Fig 1D-1G, 1H, 1K and 1L) and are concentrated in the lower part of the bristle, with almost no such Golgi satellites being seen in the distal to soma tip area. We tested for differences in the density of Golgi satellites and the relative sizes of particles in the lower part (i.e., the half of the shaft close to the soma, designated as the base) and the upper part (i.e., the half of the bristle shaft distal to the soma, designated as the tip) of the bristle shaft (see Materials and Methods for details of the quantification process used). Such analysis of MnII-GFP -positive particles revealed that the density of Golgi satellites (i.e., the number of particles per μm 2 ) at the base of the bristle shaft was significantly higher, as compared to the tip (For MnII-GFP, 0.18±0.05 and 0.04±0.06 particles/μm 2 , respectively; P<0.001, Table 1; 5 pupa, 15 bristles). In addition, the average area of each particle of both MnII-GFP was significantly different between the base and tip areas, (For MnII-GFP, 0.43±0.80 and 0.17±0.20 μm 2 , respectively; P<0.0015 Table 1; 5 pupa, 15 bristles) We further found that these Golgi satellites to be composed of cis- (Fig 1D-1G) and medial-Golgi compartments (Fig 1E, 1H and 1K) juxtaposed to ERES (Fig 1I-1L) and completely lacking trans-Golgi domains (Fig 1H and 1H'). To confirm the complete absence of trans-Golgi components within the bristle shaft, as observed in GalNacT2-RFP-expressing bristles, we tested for the presence of an additional trans-Golgi marker, 1,4-galactosyltransferase (Gal-T) [23,30]. We found no Gal-T-RFP in the wild type (WT) shaft (S1 Fig Table 1. Golgi outpost localization parameters in Dhc64C and khc mutant bristle shafts.

Dhc64C is required for Golgi organization in the bristle shaft
Given that in the Drosophila bristle, a stable MT array is organized with minus-end MT fibers pointing away from the soma [15] and with the notion that MTs and their motor proteins are required for Golgi organization [31][32][33][34], we studied the role of the minus-end MT motor protein dynein in bristle Golgi organization. First, we analyzed the role of Dhc64C using the "slow dynein" hypomorphic mutant allele Dhc64C 8-1 [17,20]. We found that in these mutants, Golgi satellites were no longer concentrated at the base of the bristle shaft as in the WT (Fig 1E-1G) but were instead found throughout the entire bristle shaft (Fig 2A-2L). These ectopically localized Golgi units contained cis- (Fig 2C, 2D, 2G and 2H), medial- (Fig 2B, 2D, 2J and 2L), and trans-Golgi compartments (Fig 2F and 2H) and ERES (Fig 2K and 2L), demonstrating that in Dhc64C mutants, the bristle shaft contains complete Golgi units rather than discrete Golgi satellites. Staining of MnII-GFP-expressing Dhc64C mutants with anti-dGM130 antibodies (a cis-Golgi marker) ( Fig 2C) revealed co-localization of cis-and medial-Golgi markers ( Fig 2D). Likewise, co-localization of GalNacT2-YFP (a trans-Golgi marker) expressed in Dhc64C mutants with anti-dGM130 antibody staining ( Fig 2F-2H) reflected co-localization of the trans-and cis-markers ( Fig 2H). Also, an additional trans-Golgi marker, 1,4-galactosyltransferase (Gal-T), was found throughout Dhc64C mutant bristle shafts (S1 Fig). We thus concluded that all three compartments are co-localized. Additionally, we showed that MnII-GFP was co-localized with ERES ( Fig 2L). Thus, the Golgi particles found within the Dhc64C mutant bristle are complete Golgi units. Quantitative analysis revealed that the density of MnII-GFP-positive particles was slightly higher at the base than at the tip (For MnII-GFP, 0.43 ±0.16 and 0.30±0.03 particles/μm 2 , respectively; P<0.05; Table 1; 5 pupa, 16 bristles). For Gal-NacT2-YFP, 0.59±0.06 and 0.24±0.02 particles/μm 2 , respectively; P<0.05; Table 2; 5 pupa, 16 bristles. In addition, the density of MnII-GFP-positive particles was dramatically increased in both the tip and base, as compared to the WT (P<0.002 for both tip and base; Table 1; 5 pupa, 16 bristles). Such analysis was not possible for GalNacT2-YFP due to its absence from the WT bristle shaft. Also, the average area of both MnII-GFP-and GalNacT2-YFP-positive particles was significantly larger at the tip of the bristle shaft than at the base (For MnII-GFP, 0.46±0.63 and 0.29±0.39 μm 2 , respectively; P<0.001; Table 1; 5 pupa, 16 bristles). For GalNacT2-YFP, 0.53±0.03 and 0.26±0.20 μm 2 , respectively; P<0.003; Table 2; 5 pupa, 16 bristles. Moreover, these areas of MnII-GFP were significantly (P<0.001) different from what was seen in the corresponding cell parts in WT flies (Table 1; this analysis was not possible for GalNacT2-YFP since the WT bristle did not contain any trans-compartments in the shaft to analyze, Table 2). The defects in Golgi organization in the Dhc64C mutants bristle shaft led us to check whether the somal region of the bristle was affected. The closer examination revealed that in Dhc64C mutants, the Golgi apparatus is scattered throughout the somal region (S2 Fig), as in WT.

Lava-lamp is not required for Golgi organization in the bristle shaft
Previously it was shown that in Drosophila embryo cellularization, Golgi movement depends on cytoplasmic dynein [35]. Moreover, it was shown that the Golgin (Golgi-associated protein) Lava-lamp (Lva) mediates such dynein-based Golgi movements [35,36]. Later, it was shown that Lva is also required for the dynamic profile of dendritic Golgi satellites [37]. Thus, we analyzed the role of Lva in bristle Golgi organization by disrupting Lva activity via over-expression of the Lva-dominant negative (Lva DN ) protein [7]. We found that expression of Lva DN in bristles using neur-Gal4 but not sca-Gal4 affected bristle morphology, with the upper part of the bristle being twisted (90% of the bristle in 10 adults; Fig 3B and 3B'). On the other hand, using the same neur-Gal4, no defects in Golgi satellite organization were detected at the bristle shaft (

Kinesin heavy chain is required for Golgi organization in the bristle shaft
Next, we examined the role of the plus-end-directed MT motor kinesin, and more specifically, the conventional kinesin-41, kinesin heavy chain (Khc), in bristle Golgi organization. Although dynein is believed to be the primary motor in Golgi positioning, it was shown in vertebrate cells that kinesin-light chain is present in Golgi membranes [38]. Still, the role of kinesin in Golgi trafficking is not well understood. Accordingly, we down-regulated the Khc transcript specifically in the bristle using RNAi (for simplicity, these strains are henceforth termed Khc mutants). To our surprise, we found that similar to Dhc64C mutant flies, bristles in Khc mutants also contained complete Golgi units located throughout the entire bristle shaft (Fig 4) composed of co-localized cis- (Fig 4C and 4D), medial- (Fig 4B, 4D, 4J and 4L), and trans-Golgi components (Fig 4F and 4H) and ERES (Fig 4K and 4L). However, the distribution pattern of the Golgi units in the Khc mutant bristle shafts was different from that in the Dhc64C mutants. Similar to WT and Dhc64C mutant flies (Tables 1 and 2), the density of MnII-GFP was significantly higher at the base of the bristle than at the tip (For MnII-GFP, 0.31±0.26 particle/μm 2 and 0.12±0.13 particle/μm 2 , respectively (P<0.001); Table 1; 5 pupa, 15 bristles). Moreover, the density of MnII-positive particles at both the tip and base was significantly higher than recorded for the WT (

Mutation of Ik2 affects Golgi organization in the bristle shaft
We previously demonstrated that in the bristle, stable MTs are organized in an asymmetrical manner, with the minus-end pointing outwards [15]. Moreover, we found that the Spn-F/Ik2 complex is required for bristle MT polarity [39]. Thus, to test whether MT polarity is essential for bristle Golgi organization, we studied Golgi organization in flies expressing Ik2 dominant-negative (Ik2 DN ) specifically in the bristle. Ik2 DN is a defective kinase in which lysine at position 41 was changed to alanine (Ik2 K41A ). We and other have shown that expression of Ik2 DN strongly affects bristle morphology and MT organization [15,40,41]. First, we found that expression with neur-Gal4 but not sca-Gal4 generated typical Ik2 bristle defects. Thus, we used the neur-Gal4 for further examinations. We found that similar to Dhc64C and Khc mutant flies, in Ik2 DN mutants, the bristle contained complete Golgi units localized throughout the entire bristle shaft (Fig 5A- 5L (and co-localized cis- (Fig 5C, 5D, 5G and 5H), medial- (Fig 5B, 5D, 5J and 5L) and trans-Golgi components (Fig 5F and 5H) and ERES (Fig 5K and 5L). Since we used different Gal4s for Golgi expression in Dhc64C and Khc-RNAi (sca-Gal4), in comparison to Ik2 DN expression (neur-Gal4), we were not able to statistically compare Golgi parameter (area and density) between them. Instead, we compared the neur-Gal4::UAS-MnII-GFP and GalNacT2-positive particle densities in Ik2 DN mutant bristle shafts versus their Gal4-UAS genetic background (neur-Gal4::UAS-MnII-GFP and GalNacT2-YFP). First, we noticed that there was no difference in MnII-GFP-positive particle average area between the tip (0.32±0.23 μm 2 , Table 3) and base (0.32±0.28 μm 2 , Table 3) in the Ik2 mutant bristle shaft. The same pattern was also detected in GalNacT2-positive particle average area (tip: 0.24±0.04 μm 2 ; base: 0.24±0.09 μm 2 , Table 3) from the Ik2 mutant. Interestingly, MnII-GFP-positive particles density was higher at the tip (0.26±0.10 particle/μm 2 , P<0.001; 5 pupa, 18 bristles, Table 3) but was similar at the base to what was noted in the WT (Table 3). In case of GalNacT2-YFP, this analysis was impossible because of its absence (Table 3). Moreover, in Ik2 DN mutants no defects in Golgi organization at the somal region of the bristle could be detected (S6 Fig).

Differential effects on Golgi organization among Dhc, Khc and Ik2 mutants
To understand the molecular mechanism by which Dhc64C, Khc and Ik2 DN affect bristle Golgi organization; we followed the timing of Golgi organization in these mutants during bristle development. Since we used a different Gal4 for Ik2 DN (neur-Gal4) than used with the Dhc64C and Khc mutants (sca-Gal4), we could only compare between Dhc64C and Khc mutants, with Ik2 DN being compared to the control. We found that at early stages of WT bristle development,  i.e. until 30-35 h APF (After prepupa formation), no Golgi satellites were detected in the bristle shaft in the WT (Fig 6A-6B). However, in both the Dhc64C (Fig 6D and 6E) and Khc ( Fig  6G and 6H) mutants, even in the early stages of elongation, e.g.,~32 h APF, MnII-GFP positive particles could be detected throughout the bristle shaft. As mentioned above, later in the development of the Dhc64C and Khc mutants, the Golgi were spread all over the bristle shaft (Fig 6F and 6I). In Ik2 DN , the timing of MnII-GFP entry into the bristle shaft was at 38 h APF in bristle development (Fig 7F).

Golgi organization in Drosophila bristles
In different Drosophila tissues, the Golgi shows different organizational patterns, reflecting differences in cell type. In most Drosophila cells, the Golgi is organized as discontinuous stacks composed of ERES and cis-, medial-, and trans-Golgi compartments [5]. During spermatogenesis, the Golgi-based acroblast has a ring shape [42,43]. In Drosophila larval neurons, the Golgi is differently organized. Here, the soma contains a high-ordered ring-shaped complete Golgi structure. In addition, similar to mammals, the dendrites contain either a discrete "single compartment Golgi" or "Golgi mini-stacks" composed of more than one Golgi compartment [23]. In our study, we showed that bristle cells contain complete Golgi stacks at the soma and incomplete Golgi composed of ERES, and cis-, and medial-Golgi compartments at the shaft. We show for the first time specifically in Drosophila that Golgi satellites can be  found in cell types other than neurons. In the bristle, Golgi satellites are preferentially localized to the shaft base area. The role of these Golgi satellites remains, however, to be determined. In dendrites, Golgi outposts localize preferentially to branch points, where they are believed to serve as MTOCs (Microtubule organizing centers) [11,44]. Thus, exact positioning of Golgi outposts is essential for establishing dendritic morphology. Additionally, in mammal neurons, Golgi outposts are more prominent during periods of intensive dendrite growth, while in mature neurons, their frequency decreases with time [14]. This points to not only tight spatial control of Golgi positioning but also to the developmental importance of the process. Here, we focused our analysis on the molecular machinery that regulates organelle positioning, although the function of Golgi satellites in the bristle cell remains to be elucidated.

The role of MTs in bristle Golgi organization
MTs are the main tracks for long-distance transport in the cell. Thus, MT array integrity and dynamics are essential for proper cellular elongation of both bristles and neurons. In highly polarized neurons, the dendrite and axon differ in terms of their MT array polarity. MTs in the axons of mammal and Drosophila neurons are oriented with their plus-ends pointed distally, while Drosophila dendrites possess MTs mainly oriented with their minus-ends pointed outwards. This difference in MT orientation is believed to provide directional cues for differential organelle transport [45][46][47]. To confirm this hypothesis, we decided to disrupt MT array polarity through expression of mutant Ik2 protein. We found that affecting MT polarity resulted in abnormal Golgi unit localization, together with ectopic trans-Golgi domain appearance. We thus showed that polarized MTs are essential for correct Golgi organization. Specifically, the presence of polarized MTs prevents the appearance of trans-Golgi components within the bristle shaft.

Possible roles of dynein and kinesin in bristle Golgi organization
Dynein is believed to be the major Golgi architecture effector [5,48,49]. In Drosophila dendrites, dynein was found to be responsible for dendrite-specific Golgi outpost localization and MT orientation [7,37,50]. In neurons, kinesin-1 plays both indirect and direct role in dendritic selective sorting of Golgi outposts [51]. Indirectly, kinesin depletion alters axon MT orientation, which leads to ectopic Golgi outposts' localization to axons driven by a different motor on misleading MT tracks. Directly, control of kinesin-1 auto-inhibition prevents entry of Golgi outposts into the axon [51]. We found that in Dhc64C mutant flies, bristle shaft Golgi satellites contained an ectopic trans-Golgi compartment that co-localized with cis-and medial-Golgi components. This raises the question of the role of Dhc64C in bristle Golgi organization. One explanation is that Dhc64C is required for Golgi satellite anchoring/sorting. In this model, dynein may prevent the entry of Golgi units from the soma to the shaft, while in the shaft, dynein specifically anchors the organelle to the lower zone close to soma, preventing the localization of Golgi units to the distant growth area of the cell tip.
A second explanation for the defects in Golgi unit localization observed in Dhc64C mutants could be disrupted interactions with the oppositely-directed motor, kinesin. However, taking into consideration MT array organization (i.e., with the minus-end pointed towards the tip) in a fly expressing a dynein mutant, active kinesin would be expected to return Golgi back to the soma. Yet, exactly the opposite is true. Moreover, we demonstrate that, similarly to Dhc64C flies, the bristle shaft contained an ectopic trans-Golgi compartment co-localized with cis-and medial Golgi components. In summary, our results shed new light of Golgi organization in polarized cells, such as the Drosophila bristle. Still, further study is needed to understand the molecular mechanism by which the combination of a polarized MT network with MT motor protein function affects Golgi bristle organization.